The demand for health care increasingly exceeds available resources and the pressure to contain the cost of medical care in the face of an ageing population is leading to growing pressure to use available resources as efficiently as possible.1In the intensive care setting, resource conservation requires careful determination of who receives treatment based on expected benefit. Intensive care triage requires a determination of which patients are ‘too well' to require intensive care2 as well as which patients are ‘too sick' to benefit.3 Considerations of individual patient's wishes and quality of life are also important.4,5Triage decisions are difficult because even the best available prediction scores are unable to predict survival with sufficient accuracy to guide decisions about individual patients and factors such as lead time bias and casemix may further confound their interpretation.6Although studies investigating the process of intensive care triage have previously been performed in the United Kingdom,7 Europe,8,9 and the United States,10 no such study has previously been conducted in Australia or New Zealand. Previous studies have demonstrated that there are significant differences in both the manner and frequency with which intensive care treatment is withheld in different parts of the world and in different cultures.11Although New Zealand and Australia are often considered culturally similar, there are significant differences in our respective cultural heritages. For example, New Zealand has larger proportions than Australia of indigenous people (15% vs 2.5%) and Polynesians (7% vs 0.5%).12-14 In addition, there are a number of factors in Australia such as greater availability of private intensive care, a higher level of funding for medical care and, a larger number of intensive care beds that are likely to lead to differences in intensive care triage practice.We hypothesised that there would be differences between Australia and New Zealand in the attitudes towards and practice of intensive care triage and sought to demonstrate these differences by way of a survey.Methods This study involved an online survey of intensive care specialists and trainees in Australia and New Zealand. The sample of 731 intensive care specialists and trainees were identified by a combination of an internet search of State (and New Zealand) Medical Registers, and a search of intensive care unit websites, supplemented by direct phone contact with individual intensive care units. The survey was designed to examine views about a range of triage scenarios and also to review recent triage practice. Prior to distribution, the survey was piloted on a small number of intensive care specialists and some minor modifications to the survey were made as a result of this. The intensive care specialists and trainees in the sample were invited to participate in the survey by an email sent on 19/05/09 that contained a link to an online survey on the website www.surveymonkey.com. A subsequent reminder was sent on 27/05/09 in order to maximise the response rate. The survey was closed to responses on 04/06/09. The survey was approved by the Multi-Region Ethics Committee of the Health Research Council of New Zealand (MEC/09/25/EXP). Respondents were required to answer a number of demographic questions including location of practice, qualifications and experience. Those respondents who made final decisions about whether or not to admit a patient were asked about recent triage decisions including the number of such decisions they had made in the previous week and how the decisions were made and communicated. All respondents were asked to consider specific triage scenarios. For each scenario they were asked to consider (a) how appropriate (on a 5-point Likert scale) they felt that intensive care was and (b) whether they would admit the patient on the basis of the information provided. The results were analysed using the R statistical package.15 Independent samples t-tests, with two-sided p-values, were used to compare means between New Zealand and Australia. Chi-squared tests were used to compare the proportions over categories, including the distribution over the 5-point Likert scale responses. Where 20% or more of the expected counts in a contingency table were less than 5 so that the standard Chi-squared test is invalid, a Monte Carlo simulation method was used to compute p-values.16, 17 In addition to Chi-square tests of association, logistic regressions were carried out to test for the importance of the explanatory variables: ‘is a trainee' (yes/no) and ‘has an additional qualification' (yes/no). Results Respondents—A total of 238 responses to the survey were obtained, representing a response rate of 33%. The proportion of respondents from various geographical locations was similar to that of all trainees and specialists registered with the Joint Faculty of Intensive Care Medicine except that the proportion of total respondents from New Zealand was higher and the proportion from New South Wales lower, than the proportion of trainees and specialists from New Zealand. The demographics of respondents are shown in Table 1. Table 1. Demographics of respondents Variables Number of respondents (%) Location of respondents New Zealand Australian Capital Territory Victoria New South Wales Queensland South Australia Northern Territory Tasmania Western Australia Other 47 (19.7%) 2 (0.8%) 46 (19.3%) 49 (20.6%) 37 (15.5%) 23 (9.7%) 6 (2.5%) 3 (1.3%) 20 (8.4%) 5 (2.1%) Respondent experience* Trainee Specialist with <5 years experience Specialist with 5-10 years experience Specialist with >10 years experience 129 (54.7%) 35 (14.8%) 24 (10.2%) 48 (20.3%) Nature of practice Tertiary or quaternary adult or mixed ICU (public) Other urban adult or mixed ICU (public) Rural ICU (public) Specialised paediatric ICU Private ICU I do not currently work in an ICU 167 (70.5%) 23 (9.7%) 7 (3.0%) 12 (5.1%) 28 (11.8%) 37 (15.6%) Additional qualifications FANZCA FACEM FRACP 47 (19.7%) 19 (8.0%) 38 (16.0%) * Two respondents did not answer this question. Recent triage decisions—Of those surveyed, 134 personally made decisions about whether to admit patients to the intensive care unit of which 77% were consultants and the remainder were trainees. These 134 were asked about triage decisions they had made in the previous week. The mean number of decisions about whether to admit a patient to the intensive care unit in New Zealand was 6.3 (95% CI 4.6-8.0) and 8.5 in Australia (95%CI 6.6-10.4) (test for difference in means, p=0.08). The mean refusal rate for the week prior to the survey was 31% (95%CI 20-42) among New Zealand respondents and 25% (95% CI 20-30) among Australian respondents (test for difference in proportions, p=0.35). 79 respondents had refused an admission in the previous week. The sources of views sought in coming to this decision are shown in Figure 1 and the means of communicating decisions to refuse admission are shown in Figure 2. No differences between Australia and New Zealand were found in the distribution of levels of experience or proportions having non-intensive care unit specialist qualifications. 41% of patients who were declined admission to intensive care were physically reviewed by an intensive care specialist. Intensive care triage scenarios—For all 7 triage scenarios described, (see Table 2) two comparisons were made between New Zealand and Australia. The first comparison was with respect to whether the respondents agreed that admission was appropriate, and the second was whether the respondents would actually admit the patient (the results are in Table 3). We also test for associations with being a trainee, and with having an additional qualification. No associations were found with respect to having additional qualifications, but in two scenarios trainees responded differently to specialists. In most of the scenarios there was a wide variety of opinions, in several cases spanning the full range of the 5-point Likert scale used to measure the strength of agreement that the patient should be admitted. Table 2. Triage scenarios A previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist describes as 'non-survivable' and who you feel may progress to brain death over the next 48 hours. A 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who has ARDS and who you believe will imminently require intubation. A 30-year-old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past five years following a severe traumatic brain injury. A 30-year-old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past five years following a severe traumatic brain injury, and has developed aspiration pneumonia due to malposition of a feeding tube in the right main bronchus. A 70-year-old male with an infective exacerbation of COPD who has a documented FEV1 of 0.9L and has failed to improve despite non-invasive ventilation for 6 hours in the emergency department. He lives at home but requires help with showering, shopping and meals. A 95-year-old man with no previous medical history who the anaesthetist, despite appropriate attempts, has been unable to extubate due to drowsiness and hypoventilation following an elective laparoscopic hernia repair. An elderly patient with a massive stroke who has been intubated in a crowded Emergency Department but is now to be extubated and palliated. The responses for all triage scenarios were similar for all States in Australia; however, the responses from New Zealand were often significantly different from those from Australia. The most striking difference was in Scenario 1, involving a previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist described as 'non-survivable' and who it was felt may progress to brain death over the next 48 hours (see Figure 3). Table 3. Results of Chi-squared tests for differences between New Zealand and Australia Scenario Agree or Strongly Agree with admission Would admit the patient Statistic p-value Statistic p-value 1 16.35 0.0001* 14.32 0.0002* 2 12.26 0.0005* 5.68 0.0280* 3 1.28 0.5767 2.66 0.2034 4 4.5 0.0339* 10.6 0.0011* 5 3.55 0.0595 12.58 0.0004* 6 1.39 0.3128 0.11 1.0000 7 0.24 0.6216 0.01 1.0000 [All tests are Chi-squared tests of association. *=significant at the 5% level] Figure 3. Scenario 1: Comparison of responses to the question: ‘Do you agree that Intensive Care Admission would be appropriate for a previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist describes as non-survivable and who you feel may progress to brain death over the next 48 hours?' and whether respondents would admit this patient Australian respondents were more positive about the appropriateness of admission in this scenario than their New Zealand counterparts (p=0.0001) and a greater proportion of Australian respondents would have admitted in this scenario (p=0.0002). Scenario 2, which described a 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who had acute respiratory distress syndrome and imminently required intubation, again demonstrated significant differences between Australia and New Zealand with respect to views on appropriateness (p=0.0005) and whether or not the patient should be admitted (p=0.03) (see Figure 4). For this Scenario, trainees were more likely than specialists to agree that admission was appropriate (p=0.0002) and more likely to say that they would admit (p=0.0023). Figure 4. Scenario 2: Comparison of responses to the question: ‘do you agree that Intensive Care Admission is appropriate for a 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who has ARDS and who you believe will imminently require intubation?' and whether respondents would admit this patient Scenarios 3 and 4 described a young patient who had been in a persistent vegetative state for five years and had pneumonia (see Figures 5 and 6). The two scenarios were the same except that in the second scenario the pneumonia was the event of an iatrogenic complication. In Scenario 3, there were no differences between the New Zealand and Australian respondents (p=0.58): both groups in general opposing admission. However, when iatrogenesis was the cause of the pneumonia a pronounced difference appeared. While New Zealand respondents generally remained strongly opposed to intensive care admission, Australian respondents were significantly less so (p=0.034); more than 20% of Australian respondents would have admitted the patient compared to 2% (just one respondent) from New Zealand (p=0.0011). In the latter Scenario, trainees were both more likely to feel that admission was appropriate (p=0.0035), and more likely to say that they would admit the patient (p=0.0071). Scenario 5 described a patient with an infective exacerbation of COPD on a background of functional impairment. There was no significant difference between New Zealand and Australian respondents in the proportions agreeing that the patients should be admitted (p=0.060). However, a significantly higher proportion of Australian respondents stated that they would in fact admit the patient (p=0.0004). In both groups the proportion that felt there was insufficient information to make a decision was high (see Figure 7). Figure 5. Scenario 3: Responses to the question: ‘Do you agree that Intensive Care admission would be appropriate for a 30 year old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past 5 years following a severe traumatic brain injury?' and whether respondents would admit this patient There was broad consensus about Scenarios 6 and 7. Scenario 6 described a 95 year old man with no previous medical history who the anaesthetist, despite appropriate attempts, had been unable to extubate due to drowsiness and hypoventilation following an elective laparoscopic hernia repair (see Figure 8). The overwhelming majority of respondents felt that intensive care admission was appropriate in the scenario described. Figure 6. Scenario 4: Responses to the question: ‘Do you agree that Intensive Care admission would be appropriate if the patient in a persistent vegetative state described in the previous question had developed aspiration pneumonia due to malposition of a feeding tube in the right main bronchus?' and whether respondents would admit this patient

The demand for health care increasingly exceeds available resources and the pressure to contain the cost of medical care in the face of an ageing population is leading to growing pressure to use available resources as efficiently as possible.1In the intensive care setting, resource conservation requires careful determination of who receives treatment based on expected benefit. Intensive care triage requires a determination of which patients are ‘too well' to require intensive care2 as well as which patients are ‘too sick' to benefit.3 Considerations of individual patient's wishes and quality of life are also important.4,5Triage decisions are difficult because even the best available prediction scores are unable to predict survival with sufficient accuracy to guide decisions about individual patients and factors such as lead time bias and casemix may further confound their interpretation.6Although studies investigating the process of intensive care triage have previously been performed in the United Kingdom,7 Europe,8,9 and the United States,10 no such study has previously been conducted in Australia or New Zealand. Previous studies have demonstrated that there are significant differences in both the manner and frequency with which intensive care treatment is withheld in different parts of the world and in different cultures.11Although New Zealand and Australia are often considered culturally similar, there are significant differences in our respective cultural heritages. For example, New Zealand has larger proportions than Australia of indigenous people (15% vs 2.5%) and Polynesians (7% vs 0.5%).12-14 In addition, there are a number of factors in Australia such as greater availability of private intensive care, a higher level of funding for medical care and, a larger number of intensive care beds that are likely to lead to differences in intensive care triage practice.We hypothesised that there would be differences between Australia and New Zealand in the attitudes towards and practice of intensive care triage and sought to demonstrate these differences by way of a survey.Methods This study involved an online survey of intensive care specialists and trainees in Australia and New Zealand. The sample of 731 intensive care specialists and trainees were identified by a combination of an internet search of State (and New Zealand) Medical Registers, and a search of intensive care unit websites, supplemented by direct phone contact with individual intensive care units. The survey was designed to examine views about a range of triage scenarios and also to review recent triage practice. Prior to distribution, the survey was piloted on a small number of intensive care specialists and some minor modifications to the survey were made as a result of this. The intensive care specialists and trainees in the sample were invited to participate in the survey by an email sent on 19/05/09 that contained a link to an online survey on the website www.surveymonkey.com. A subsequent reminder was sent on 27/05/09 in order to maximise the response rate. The survey was closed to responses on 04/06/09. The survey was approved by the Multi-Region Ethics Committee of the Health Research Council of New Zealand (MEC/09/25/EXP). Respondents were required to answer a number of demographic questions including location of practice, qualifications and experience. Those respondents who made final decisions about whether or not to admit a patient were asked about recent triage decisions including the number of such decisions they had made in the previous week and how the decisions were made and communicated. All respondents were asked to consider specific triage scenarios. For each scenario they were asked to consider (a) how appropriate (on a 5-point Likert scale) they felt that intensive care was and (b) whether they would admit the patient on the basis of the information provided. The results were analysed using the R statistical package.15 Independent samples t-tests, with two-sided p-values, were used to compare means between New Zealand and Australia. Chi-squared tests were used to compare the proportions over categories, including the distribution over the 5-point Likert scale responses. Where 20% or more of the expected counts in a contingency table were less than 5 so that the standard Chi-squared test is invalid, a Monte Carlo simulation method was used to compute p-values.16, 17 In addition to Chi-square tests of association, logistic regressions were carried out to test for the importance of the explanatory variables: ‘is a trainee' (yes/no) and ‘has an additional qualification' (yes/no). Results Respondents—A total of 238 responses to the survey were obtained, representing a response rate of 33%. The proportion of respondents from various geographical locations was similar to that of all trainees and specialists registered with the Joint Faculty of Intensive Care Medicine except that the proportion of total respondents from New Zealand was higher and the proportion from New South Wales lower, than the proportion of trainees and specialists from New Zealand. The demographics of respondents are shown in Table 1. Table 1. Demographics of respondents Variables Number of respondents (%) Location of respondents New Zealand Australian Capital Territory Victoria New South Wales Queensland South Australia Northern Territory Tasmania Western Australia Other 47 (19.7%) 2 (0.8%) 46 (19.3%) 49 (20.6%) 37 (15.5%) 23 (9.7%) 6 (2.5%) 3 (1.3%) 20 (8.4%) 5 (2.1%) Respondent experience* Trainee Specialist with <5 years experience Specialist with 5-10 years experience Specialist with >10 years experience 129 (54.7%) 35 (14.8%) 24 (10.2%) 48 (20.3%) Nature of practice Tertiary or quaternary adult or mixed ICU (public) Other urban adult or mixed ICU (public) Rural ICU (public) Specialised paediatric ICU Private ICU I do not currently work in an ICU 167 (70.5%) 23 (9.7%) 7 (3.0%) 12 (5.1%) 28 (11.8%) 37 (15.6%) Additional qualifications FANZCA FACEM FRACP 47 (19.7%) 19 (8.0%) 38 (16.0%) * Two respondents did not answer this question. Recent triage decisions—Of those surveyed, 134 personally made decisions about whether to admit patients to the intensive care unit of which 77% were consultants and the remainder were trainees. These 134 were asked about triage decisions they had made in the previous week. The mean number of decisions about whether to admit a patient to the intensive care unit in New Zealand was 6.3 (95% CI 4.6-8.0) and 8.5 in Australia (95%CI 6.6-10.4) (test for difference in means, p=0.08). The mean refusal rate for the week prior to the survey was 31% (95%CI 20-42) among New Zealand respondents and 25% (95% CI 20-30) among Australian respondents (test for difference in proportions, p=0.35). 79 respondents had refused an admission in the previous week. The sources of views sought in coming to this decision are shown in Figure 1 and the means of communicating decisions to refuse admission are shown in Figure 2. No differences between Australia and New Zealand were found in the distribution of levels of experience or proportions having non-intensive care unit specialist qualifications. 41% of patients who were declined admission to intensive care were physically reviewed by an intensive care specialist. Intensive care triage scenarios—For all 7 triage scenarios described, (see Table 2) two comparisons were made between New Zealand and Australia. The first comparison was with respect to whether the respondents agreed that admission was appropriate, and the second was whether the respondents would actually admit the patient (the results are in Table 3). We also test for associations with being a trainee, and with having an additional qualification. No associations were found with respect to having additional qualifications, but in two scenarios trainees responded differently to specialists. In most of the scenarios there was a wide variety of opinions, in several cases spanning the full range of the 5-point Likert scale used to measure the strength of agreement that the patient should be admitted. Table 2. Triage scenarios A previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist describes as 'non-survivable' and who you feel may progress to brain death over the next 48 hours. A 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who has ARDS and who you believe will imminently require intubation. A 30-year-old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past five years following a severe traumatic brain injury. A 30-year-old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past five years following a severe traumatic brain injury, and has developed aspiration pneumonia due to malposition of a feeding tube in the right main bronchus. A 70-year-old male with an infective exacerbation of COPD who has a documented FEV1 of 0.9L and has failed to improve despite non-invasive ventilation for 6 hours in the emergency department. He lives at home but requires help with showering, shopping and meals. A 95-year-old man with no previous medical history who the anaesthetist, despite appropriate attempts, has been unable to extubate due to drowsiness and hypoventilation following an elective laparoscopic hernia repair. An elderly patient with a massive stroke who has been intubated in a crowded Emergency Department but is now to be extubated and palliated. The responses for all triage scenarios were similar for all States in Australia; however, the responses from New Zealand were often significantly different from those from Australia. The most striking difference was in Scenario 1, involving a previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist described as 'non-survivable' and who it was felt may progress to brain death over the next 48 hours (see Figure 3). Table 3. Results of Chi-squared tests for differences between New Zealand and Australia Scenario Agree or Strongly Agree with admission Would admit the patient Statistic p-value Statistic p-value 1 16.35 0.0001* 14.32 0.0002* 2 12.26 0.0005* 5.68 0.0280* 3 1.28 0.5767 2.66 0.2034 4 4.5 0.0339* 10.6 0.0011* 5 3.55 0.0595 12.58 0.0004* 6 1.39 0.3128 0.11 1.0000 7 0.24 0.6216 0.01 1.0000 [All tests are Chi-squared tests of association. *=significant at the 5% level] Figure 3. Scenario 1: Comparison of responses to the question: ‘Do you agree that Intensive Care Admission would be appropriate for a previously well 50-year-old woman with an intracerebral haemorrhage that the neurologist describes as non-survivable and who you feel may progress to brain death over the next 48 hours?' and whether respondents would admit this patient Australian respondents were more positive about the appropriateness of admission in this scenario than their New Zealand counterparts (p=0.0001) and a greater proportion of Australian respondents would have admitted in this scenario (p=0.0002). Scenario 2, which described a 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who had acute respiratory distress syndrome and imminently required intubation, again demonstrated significant differences between Australia and New Zealand with respect to views on appropriateness (p=0.0005) and whether or not the patient should be admitted (p=0.03) (see Figure 4). For this Scenario, trainees were more likely than specialists to agree that admission was appropriate (p=0.0002) and more likely to say that they would admit (p=0.0023). Figure 4. Scenario 2: Comparison of responses to the question: ‘do you agree that Intensive Care Admission is appropriate for a 30-year-old female with relapsed acute myeloid leukaemia following a bone marrow transplant who has ARDS and who you believe will imminently require intubation?' and whether respondents would admit this patient Scenarios 3 and 4 described a young patient who had been in a persistent vegetative state for five years and had pneumonia (see Figures 5 and 6). The two scenarios were the same except that in the second scenario the pneumonia was the event of an iatrogenic complication. In Scenario 3, there were no differences between the New Zealand and Australian respondents (p=0.58): both groups in general opposing admission. However, when iatrogenesis was the cause of the pneumonia a pronounced difference appeared. While New Zealand respondents generally remained strongly opposed to intensive care admission, Australian respondents were significantly less so (p=0.034); more than 20% of Australian respondents would have admitted the patient compared to 2% (just one respondent) from New Zealand (p=0.0011). In the latter Scenario, trainees were both more likely to feel that admission was appropriate (p=0.0035), and more likely to say that they would admit the patient (p=0.0071). Scenario 5 described a patient with an infective exacerbation of COPD on a background of functional impairment. There was no significant difference between New Zealand and Australian respondents in the proportions agreeing that the patients should be admitted (p=0.060). However, a significantly higher proportion of Australian respondents stated that they would in fact admit the patient (p=0.0004). In both groups the proportion that felt there was insufficient information to make a decision was high (see Figure 7). Figure 5. Scenario 3: Responses to the question: ‘Do you agree that Intensive Care admission would be appropriate for a 30 year old male with pneumonia requiring ventilation who has been in a nursing home in a persistent vegetative state for the past 5 years following a severe traumatic brain injury?' and whether respondents would admit this patient There was broad consensus about Scenarios 6 and 7. Scenario 6 described a 95 year old man with no previous medical history who the anaesthetist, despite appropriate attempts, had been unable to extubate due to drowsiness and hypoventilation following an elective laparoscopic hernia repair (see Figure 8). The overwhelming majority of respondents felt that intensive care admission was appropriate in the scenario described. Figure 6. Scenario 4: Responses to the question: ‘Do you agree that Intensive Care admission would be appropriate if the patient in a persistent vegetative state described in the previous question had developed aspiration pneumonia due to malposition of a feeding tube in the right main bronchus?' and whether respondents would admit this patient